Introduction & Overview

Welcome to the Kla Digital Developer Documentation. This guide is designed for developers integrating Kla's AgentOps platform into their own systems.

What is Kla Digital?

Kla Digital provides a comprehensive AgentOps Platform that enables enterprises to run, audit, and govern AI agents with confidence. Our platform bridges the gap between experimental AI and production-grade reliability by providing:

  • Execution: A robust runtime for AI agents, powered by Temporal workflows.
  • Audit: Immutable, cryptographically verifiable logs of every action and decision.
  • Governance: Policy-as-code enforcement to ensure compliance with regulations like the EU AI Act, SOC 2, and ISO 27001.

Key Concepts

🤖 Agents & Manifests

An Agent is a unit of AI functionality defined by a Manifest. The manifest describes the agent's capabilities, tools, LLM configuration, and resource requirements. Manifests are content-addressable (referenced by SHA-256 hash) to ensure reproducibility.

⚡ Executions

An Execution is a single run of an agent. It is a stateful process that can be long-running, paused for human approval, and resumed. Executions are orchestrated by the Execution API and run on durable workers.

📜 Policies

Policies are rules defined in Cerbos that govern what agents can and cannot do. Policies are evaluated at runtime for every significant action, ensuring that agents operate within safe bounds.

🔍 Audit Trail

The Audit Trail is an immutable ledger (backed by ImmuDB) that records every execution step, policy decision, and system event. This trail provides the "proof of work" needed for compliance audits.

Getting Started

To start integrating with Kla Digital, we recommend following the Quickstart Guide to run your first agent.

Integration Points

Authentication

All API endpoints require a valid JWT Token. You can obtain a token via your tenant's OIDC provider (e.g., Keycloak).

Authorization: Bearer <your-jwt-token>